Proper unicode font support is like 1.2GiB (noto, but I haven't found any complete unicode font collections that are significantly smaller). There's bloat for sure, but supporting universal text is one that I think is not a waste of space.
Maybe not proper support, but when I tried NetBSD recently my entire installation was around 1.5 GB on disk and seemed to handle Unicode well enough for me (for languages I care about). Not doubting some more packages would be needed to support every language, but happy everything wasn't installed by default.
For English capitalization is a trivial problem. I think for Hungarian or something similar the rule set is like 6mb.
Unsure if this is useful to you but have you heard about GNU Unifont? It’s not as nice and comes with some asterisks but damn it’s very compact.
I first read about it via this blog post: https://shkspr.mobi/blog/2019/04/banish-the-%ef%bf%bd-with-u...